Internat. J. Math. & Math. Sci. Vol. 23, No. 3 (2000) 217–222 S0161171200001198 © Hindawi Publishing Corp. CONVEX ISOMETRIC FOLDING E. M. ELKHOLY (Received 27 September 1993 and in revised form 5 June 1996) Abstract. We introduce a new type of isometric folding called “convex isometric folding.” We prove that the infimum of the ratio Vol N/ Vol ϕ(N) over all convex isometric foldings ϕ : N → N, where N is a compact 2-manifold (orientable or not), is 1/4. Keywords and phrases. Isometric and convex foldings, regular and universal covering spaces, manifolds, group of covering transformations, fundamental regions. 2000 Mathematics Subject Classification. Primary 51H10, 57N10. 1. Introduction. A map ϕ : M → N, where M and N are C ∞ Riemannian manifolds of dimensions m and n, respectively, is said to be an isometric folding of M into N if and only if for any piecewise geodesic path γ : J → M, the induced path ϕ ◦γ : J → N is a piecewise geodesic and of the same length. The definition is given by Robertson [4]. The set of all isometric foldings ϕ : M → N is denoted by (M, N). Let p : M → N be a regular locally isometric covering and let G be the group of covering transformations of p. An isometric folding φ ∈ (M) is said to be p-invariant if and only if for all g ∈ G and all x ∈ X, p(ϕ(x)) = p(ϕ(g, x)). See Robertson and Elkholy [5]. The set of p-invariant isometric foldings is denoted by i (M, p). Definition 1.1. Let ϕ ∈ (M, N), where M and N are C ∞ Riemannian manifolds of dimensions m and n, respectively. We say that ϕ is a convex isometric folding if and only if ϕ(M) can be embedded as a convex set in Rn . We denote the set of all convex isometric foldings of M into N by C(M, N), and if C(M, N) ≠ , then it forms a subsemigroup of (M, N). Definition 1.2. We say that ϕ ∈ i (M, p) is a p-invariant convex isometric folding if and only if ϕ(M) can be embedded as a convex set in Rm . We denote the set of p-invariant convex isometric foldings of M by Ci (M, p). If Ci (M, p) ≠ , then for any covering map, p : M → N, Ci (M, p) is a subsemigroup of C(M). To solve our main problem we need the following: (1) Robertson and Elkholy [5] proved that if N is an n-smooth Riemannian manifold, p : M → N is its universal covering, and G is the group of covering transformations of p, then (N) is isomorphic as a semigroup to i (M, p)/G. (2) Elkholy [1] proved that if N is an n-smooth Riemannian manifold, p : M → N is its universal covering, and ϕ ∈ (N) such that ϕ∗ : π1 (N) → π1 (N) is trivial, then the 218 E. M. ELKHOLY corresponding folding ψ ∈ i (M, p) maps each fiber of p to a single point. (3) Elkholy and Al-Ahmady [3] proved that under the same conditions of (2), if N is a compact 2-manifold, then Vol F Vol N = , Vol ϕ(N) Vol ψ(F ) (1.1) where F is a fundamental region of G in M. 2. Convex isometric folding and covering spaces. The next theorem establishes the relation between the set of convex isometric folding of a manifold, C(N), and the set of p-invariant convex isometric folding of its universal covering space, Ci (M, p). Theorem 2.1. Let N be a manifold and p : M → N its universal covering. Let G be the group of covering transformations of p. If C(N) ≠ , then C(N) is isometric as a semigroup to Ci (M, p)/G. Proof. Let C(N) ≠ ϕ. Then by using (1), there exists an isomorphism f from i (M, p)/G into (N). Since Ci (M, p) is a subsemigroup of i (M, p), Ci (M, p)/G is a subsemigroup of i (M, p)/G. Let h = f | (Ci (M, P )/G). Since Ci (M, p)/G is a semigroup, h is a homeomorphism and also it is one-one. To show that h is an onto map, we suppose that ϕ ∈ C(N). Hence, ϕ ∈ (N) and, consequently, there exists ψ ∈ i (M, p)/G. Since ϕ ∈ C(N), ϕ∗ is trivial and hence for all x ∈ M, ψ(G, x) = ψ(x), and therefore ψ ∈ Ci (M, p)/G. Theorem 2.2. Let N be a compact orientable 2-manifold and consider the universal covering space (R2 , P ) of N. Let ϕ ∈ C(N) and ψ ∈ Ci (R2 , p). Then for all x, y ∈ R2 , d(ψ(x), ψ(y)) ≤ ∆, where ∆ is the radius of a fundamental region for the covering space. Proof. Elkholy [1] proved the truth of the theorem for N = S 2 . So, we have to prove it for the connected sum of n-tori. First, let N = T be a torus homomorphic to the quotient space obtained by identifying opposite sides of a square of length “a” as shown in Figure 1(a) a g2 · x g3 · x b T b α3 α4 a g4 · x (b) (a) Figure 1. α2 y α1 g1 · x 219 CONVEX ISOMETRIC FOLDING Suppose that ϕ : T → T is a convex isometric folding. Then ϕ∗ (π1 (T )) is trivial. By Theorem 2.1, there exists a convex isometric folding ψ : R2 → R2 such that for all x, y ∈ R2 and for all g ∈ G, p ψ(x) = p ψ(g, x) . Equivalently, for all (P , Q) ∈ R2 and for all g ∈ Z × Z, there exists a unique h ∈ Z × Z such that h ◦ ψ(P , Q) = ψ g(P , Q) , i.e., ψ P , Q + 2∆m, 2∆n = ψ P + 2∆m, Q + 2∆n , where m, n, m , n ∈ Z. (2.1) Consider any fundamental region F of the covering space (R2 , p) of T , i.e., a closed square of length “a” with sides identified as shown in Figure 1(b). Since ϕ∗ is trivial, by (2), for all x ∈ R2 , ψ(G, x) = ψ(x). Now, let x and y be distinct points of R2 such that x = g · y for all g ∈ G and let d(x, y) = α1 . Then there exists a point x ∗ = g · x such that d y, x ∗ = min(αi ), αi = d y, gi , x , i = 1, . . . , 4. (2.2) Thus, there are always four equivalent points gi · x, i = 1, . . . , 4 which form the vertices of a square of length “a” and such that d(gi · x, y) ≤ 2∆. From Figure 1(b), it is clear that max d(x ∗ , y) ≤ ∆ and since ψ is an isometric folding, by Robertson [4], d(ψ(x), ψ(y)) ≤ d(x, y), i.e., d ψ(x), ψ y = d ψ gi · x , ψ y ≤ d gi · x, y = d x, y ≤ ∆, (2.3) and this proves the theorem for N = T . Now, consider the connected sum of two tori, obtained as a quotient space of an octagon with sides identified as shown in Figure 2(a). The group of covering transformations G is isometric to Z × Z × Z × Z. Using the same previous technique, we can g2 · x g3 · x T #T a3 a4 a2 y a1 g4 · x g1 · x (b) (a) Figure 2. obtain four equivalent points as the vertices of a square of diameter 2∆ such that max d(y, x ∗ ) ≤ ∆, and the result follows. This theorem, by using the above method, is true for the connected sum of n-tori. 220 E. M. ELKHOLY Theorem 2.3. Let N be a compact nonorientable 2-manifold and consider the universal covering space (M, p) of N. Let φ ∈ C(N) and ψ ∈ Ci (M, p). Then for all x, y ∈ M, d ψ(x), ψ(y) ≤ ∆, where ∆ is the radius of a fundamental region for the covering space. Proof. By Elkholy [2], the theorem is true for N = p 2 and M = S 2 . Now, consider the connected sum of two projective planes, the Klein bottle K, homeomorphic to the quotient space obtained by identifying the opposite sides of a square as shown in Figure 3(a). a b K g4 · x g3 · x b y g1 · x a (a) g2 · x (b) Figure 3. Suppose that ϕ : K → K is a convex isometric folding. Then there exists a convex isometric folding ψ : R2 → R2 such that for all x ∈ R2 and g ∈ G, p(ψ(x)) = p(ψ(g · x)). Equivalently, for all (P , Q) ∈ R2 and for all g ∈ Z × Z2 , there exists a unique h ∈ Z × Z2 such that h ◦ ψ(P , Q) = ψ(g(P , Q)), i.e., ψ P , Q + 2∆m , 2∆n = ψ P + 2∆m, 2∆n + (−)m Q , where m, n, m , n ∈ Z. (2.4) Any fundamental region F of the covering space (R2 , p) of K is a closed square of diameter 2∆ with the boundary identified as shown in Figure 3(b). Since ϕ∗ is trivial, for all x ∈ R 2 , ψ(G · x) = ψ(x). Now, let x and y be distinct points of R2 such that y ≠ g · x for all g ∈ G, and let d(x, y) = α1 . Thus, there exists a point x ∗ = g · x such that d y, x ∗ = min(αi ), αi = d y, gi · x , i = 1, . . . , 4. (2.5) Thus, there are always four equivalent points gi · x which form the vertices of a parallelogram such that the shortest diameter is of length less than 2∆. Now, the point y is either inside or on the boundary of a triangle of vertices g1 ·x = x, g2 · x, g3 · x. Let y be a point equidistant from the vertices of this triangle, i.e., d y , x = d y , g2 · x = d y , g3 · x . (2.6) CONVEX ISOMETRIC FOLDING 221 From Figure 3(b), it is clear that d(y , x) < ∆ and, hence, d(x ∗ , y) < ∆. Therefore, d ψ(x), ψ y = d ψ gi · x , ψ y ≤ d g · xi , y = d x ∗ , y < ∆ (2.7) and the result follows. Now, let N be the connected sum of three projective planes obtained as the quotient space of a hexagon with the sides identified in pairs as indicated in Figure 4(a). In this case, (R2 , p) is the universal cover of N and G Z × Z × Z2 . Using the same method as that used above, we can always have equivalent points gi · x, i = 1, . . . , 4 which form the vertices of a parallelogram whose shortest diameter is of length less than 2∆. From Figure 4(b), we can see that max d(y, x ∗ ) < ∆ and the theorem is proved. In general and by using the same technique, the theorem is also true for the connected sum of n-projective planes. g4 · x g3 · x y g1 · x g2 · x P #P #P (a) (b) Figure 4. 3. Volume and convex folding. The following theorem succeeds in estimating the maximum volume we may have if we convexly folded a compact 2-manifold into itself. Theorem 3.1. The infimum of the ratio eN = Vol N , Vol ϕ(N) (3.1) where N is a compact 2-manifold over all convex isometric foldings ϕ ∈ C(N) of degree zero, is 4. Proof. Robertson [4] has shown that if N is a compact 2-manifold, and ϕ : N → N is a convex isometric folding, any convex isometric folding is an isometric folding, then deg ϕ is ±1 or 0. We consider only the case for which deg ϕ is zero otherwise ϕ(N) cannot be embedded as a convex subset of R2 unless N is. In this case, the set of singularities of ϕ decomposes N into an even number of strata, say k, each of which is homeomorphic to ϕ(N) and, hence, Vol N = k Vol ϕ(N), (3.2) 222 E. M. ELKHOLY that is, eN should be an even number. To calculate the exact value of eN , consider first an orientable 2-compact manifold N. By using (1.1) eN = Vol F Vol ϕ(F ) (3.3) and this means that eN can be calculated by calculating the volume of F and of its image ϕ(F ), but F is a closed square of diameter 2∆ and ϕ(F ) is a closed subset of F such that the distance d(x, x ) between any two points x, x ∈ ϕ(F ) is at most ∆. The supremum of 2-dimensional volume of such set is φ(∆/2)2 and, hence, 2 < eN . But eN is an even number. Hence, eN = 4. Now, let N be a nonorientable 2-compact manifold, i.e., a connected sum of nprojective planes. Elkholy [2] proved the theorem for n = 1. The fundamental region in this case is a square or a rectangle of diameter 2∆ according to whether n is even or odd. If n is an even number, then Vol F = 2∆2 (3.4) and the result follows. Now, let n be an odd number. Then F is a rectangle of lengths ((n + 1)/2)a, ((n − 1)/2)a and hence a(n + 1)/2 a(n − 1)/2 n2 − 1 2∆2 . Vol F = 4∆2 sin θ cos θ = 4∆2 = 2 2 2 n +1 a (n + 1)/2 a (n + 1)/2 (3.5) Therefore, eN > 2 for all n > 1. Since eN is an even number, eN = 4. References [1] [2] [3] [4] [5] E. M. Elkholy, Isometric and topological folding of manifolds, Ph.D. thesis, Southampton University, England, 1982. , Maximum isometric folding of the real projective plane, J. Natur. Sci. Math. 25 (1985), no. 1, 25–30. MR 87d:53081. Zbl 589.53052. E. M. Elkholy and A. E. Al-Ahmady, Maximum volume of isometric foldings of compact 2Riemannian manifolds, Proc. Conf. Oper. Res. and Math. Methods Bull. Fac. Sc. Alex. 26 A (1977), no. 1, 54–66. S. A. Robertson, Isometric folding of Riemannian manifolds, Proc. Roy. Soc. Edinburgh Sect. A 79 (1977), no. 3–4, 275–284. MR 58 7486. Zbl 418.53016. S. A. Robertson and E. M. Elkholy, Invariant and equivariant isometric folding, Delta J. Sci. 8 (1984), no. 2, 374–385. Elkholy: Department of Mathematics, Faculty of Science, Tanta University, Tanta, Egypt